Upregulation of DNA Repair Genes in Active Cirrhosis Associated with Hepatocellular Carcinomaq

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

FEBS 29091 FEBS Letters 579 (2005) 95–99

Upregulation of DNA repair genes in active cirrhosis associated with hepatocellular carcinomaq

Pierre Zindya, Lise Andrieuxa, Dominique Bonniera, Orlando Mussoa, Sophie Langoue¨ta, Jean Pierre Campiona,b, Bruno Turlinc, Bruno Cle´menta, Nathalie The´reta,* a INSERM U620, De´toxication et Re´paration Tissulaire, Faculte´sdeMe´decine et Pharmacie, Universite´ de Rennes I, 2 Av. Le´on Bernard, 35043 Rennes Cedex, France b De´partement de Chirurgie Visce´rale, Hoˆpital Pontchaillou, Rennes, France c Service dÕAnatomie Pathologique, Hoˆpital Pontchaillou, Rennes, France

Received 15 July 2004; revised 2 September 2004; accepted 7 September 2004

Available online 30 November 2004

Edited by Gianni Cesareni

sis associated with HCC [5]. The present study was designed to Abstract Phenotypic changes in injured livers involve complex network of genes whose interplays may lead to fibrosis and cir- investigate gene networks involved in the response to chronic rhosis, a major risk of hepatocellular carcinoma. Gene expres- injury. Instead of analyzing this process with a gene-by-gene sion profiles in fibrotic livers were analyzed by using cDNA approach, we chose to combine cDNA microarray and a bio- microarray, hierarchical clustering and gene ontology. Analyses informatics tool that carries out simple datamining using gene of a major cluster of upregulated genes in cirrhosis identified a ontology (GO). Differentially expressed genes in microarray new set of genes involved in DNA repair and damage. The upreg- experiments evidenced a new cluster of genes associated with ulation of DNA repair genes was confirmed by real-time quanti- DNA repair, including mechanistically distinct repair systems, tative polymerase chain reaction and associated with i.e., nucleotide excision repair, homologous recombination and necroinflammatory activity (P < 0.001). Increased DNA repair post-replicative repair. In addition, our data show an associa- activity in cirrhosis with inflammatory activity may reflect in- tion between upregulated DNA repair genes and cirrhosis creased DNA damages as a consequence of chronic liver injury. 2004 Federation of European Biochemical Societies. Published activity assessed by piecemeal necrosis, lobular necrosis and by Elsevier B.V. All rights reserved. portal inflammation.

Keywords: Liver; DNA repair; Gene ontology 2. Materials and methods

2.1. Patients and tissue samples Matching non-tumor livers were obtained from patients undergoing surgical hepatectomy or liver transplantation for hepatocellular carci- 1. Introduction noma. Controls were histologically normal liver samples, obtained from the non-tumoral parts of livers complicated with metastatic tu- Most chronic liver diseases may lead to cirrhosis, which is mor. The histologic stage of fibrosis and the intensity of necroinflam- characterized by extensive connective tissue deposition, regen- matory lesions were graded according to the METAVIR score [6]: AO = no activity, A1 = mild, A2 = moderate, A3 = severe and F0, eration, vascular disorders and hepatocyte necrosis. It is the no fibrosis; F1, portal fibrosis without septa; F2, portal fibrosis with main risk factor for the development of hepatocellular carci- rare septa; F3 numerous septa without cirrhosis and F4 cirrhosis. Clin- noma (HCC), cirrhotic macronodules being suggested as po- ical characteristics of patients and histologic annotations are shown in tential premalignant lesions [1–3]. Thus, it is hypothesized Table 1. Access to this material was in agreement with French laws and satisfied the requirements of the local Ethics Committee. that the transformation of hepatocytes towards a neoplastic phenotype results from a cascade of molecular events leading 2.2. Microarray hybridizations to irreversible DNA damages [4]. Recently, large overall Fourteen samples including three F1, three F2, three F3 and five F4 screening of gene expression by cDNA array analyses has iden- stage were analyzed. Microarray hybridization (Atlas Human Cancer tified upregulated genes involved in inflammation, fibrosis and 1.2 Array, Clontech, Palo Alto, CA) was performed as indicated by regeneration, and a molecular signature including genes in- the manufacturer and each sample was compared with a pool of mRNA volved in several molecular pathways was proposed for cirrho- extracted from control livers (n = 14). Both spot radioactive quantifica- tion of intensity and global normalization using the sum method were realized using the Atlas Image 2.01 software (Clontech). The ratio of q each fibrotic sample versus the control pool sample intensity was then This work was supported by the Institut National de la Sante´ et de la calculated. The reproducibility of the microarray data was evaluated be- Recherche Me´dicale, the Association pour la Recherche contre le tween five independent hybridizations of the pool control and the corre- Cancer and the Region Bretagne (PRIR No. 139). Pierre Zindy was a lation coefficients varied from 0.64 to 0.81 (P < 0.05). recipient of a fellowship from la Ligue contre le Cancer (Ille et Vilaine).

*Corresponding author. Fax: +33 2 23234794. 2.3. Cluster, statistical and ontology analyses E-mail address: [email protected] (N. The´ret). Before analysis, gene expression data were log2-transformed and centralized. The cluster analysis was performed with average uncen- Abbreviations: HCC, hepatocellular carcinoma; RT-qPCR, reverse tered linkage clustering by using Cluster TreeView software (http:// transcription quantitative polymerase chain reaction; GO, Gene rana.Stanford.EDU/software). To select signiﬁcant genes, we used a ontology two-class unpaired t test with permutation using statistical tool

Table 1 2.6. Immunohistochemistry Clinical characteristics of patients. (\: samples analyzed by macro- Five-micrometer-thick frozen sections were fixed in 4% paraformal- array) dehyde buffered in PBS (pH 7.4), blocked in 2% BSA–PBS (w/v), incubated with antibodies at 4 C overnight and subsequently incubated Case Age Gender Alcohol HCV HBV Metavir score with peroxidase conjugated goat anti-rabbit IgG (Sigma, St. Quentin, 153M A0F1 France). Specific antibodies were omitted in controls and all sections 276M + A0F1 were counterstained with hematoxylin. 361M + A0F1 472M A1F1 2.7. P53 gene sequencing \ 561M + A0F1 DNA was amplified by PCR between exon 4 and exon 8 (sense: 50- \ 675M A0F1 tactcccctgccctcaacaa-30, antisense: 50-cagtgctcgcttagtgct-30) and direct 772M A0F1 DNA sequencing was performed with a BigDye sequencing kit (Ap- 876M A1F1 plied Biosystem, CA). Electrophoresis and sequence analysis were per- \ 973M + A0F1 formed with ABI PRISM 310. 10 73 M + A3F2 11 75 M A1F2 \ 12 72 M + + A0F2 13 62 M + A0F2 3. Results and discussion \ 14 56 M A0F2 \ 15 68 M A1F2 16 64 M + A0F2 Analyses of membranes by the Atlas Image 2.01 software 17 81 M A1F2 demonstrated that 697 plotted genes showed a significant sig- 18 68 M + A0F3 nal (>5000) in at least one of the experiments, indicating a cor- 19 66 M + A1F3 \ 20 74 M + A0F3 \ 21 66 M A0F3 22 64 M + A2F3 23 67 M + + A1F3 \ 24 63 M A0F3 \ 25 74 M A0F4 26 61 M + + A0F4 27 54 M A0F4 \ 28 52 M + A0F4 \ 29 50 M + A0F4 30 60 M + A3F4 31 52 M + + A2F4 32 67 M + A1F4 33 57 M + + A2F4 34 65 M + A2F4 35 63 M + A2F4 \ 36 37 M + A3F4 37 66 M + A2F4 38 59 M + A2F4 \ 39 56 F A1F4

SAM (http://www-stat.stanford.edu/~tibs/SAM/). In order to iden- tify the biological processes involved by modulated gene, we performed a GO analysis by using FatiGO (http:// fatigo.bioinfo.cnio.es), a web interface which carries out simple datamining using GO for DNA microarray data (http://www.geneon- tology.org/).

2.4. Real-time quantitative PCR (RT-qPCR) Real-time quantitative PCR was performed on RNA extracted from tissue samples with the fluorescent dye SYBR Green methodology using the qPCRe Core Kit for Sybre Green I from Eurogentec (Ser- aing, Belgium) and the ABI Prism 7000 (Perkin–Elmer, Foster City, CA). Primer pairs were: P53, sense: 50-tgcaataggtgtgcgtcagaa-30, antisense: 50-tgcaataggtgtgcgtcagaa-30, ATM, sense: 50-tgacgttacatgagccagca-30, antisense: 50-tgacgttacatgagccagca-30; RAD23A, sense: 50- ccaggagaaccctcagcttt-30, antisense: 50-tcagcatctggatgaactgc-30; UBE2A, sense: 50-ggagcagaggaatggaaaca-30, antisense: 50-agggcagcact- gatccttta-30; ERCC1, sense: 50-tagcggaggctgaggaaca-30, antisense: 50- ggcgacgtaattcccgact-30; XPA, sense: 50-gcaccactgtaccccaggtt-30, antisense: 50-ttccccactgtttccacca-30; XRCC1, sense: 50-cagccctacagcaagg- 0 0 0 Fig. 1. Hierarchical clustering of 14 patients with liver fibrosis and actc-3 , antisense: 5 -gcctctgcctcatctttgtc-3 . selected 697 genes analyzed using Cluster and TreeView software (http://rana.Stanford.EDU/software/). (A) Red and green colors illus- 2.5. Western blot analyses trate over- and under-expressed genes in fibrotic tissues, respectively. Frozen serial sections from tissue blocks were scrapped and proteins Black and gray represent unchanged or absent gene expression. (B) solubilized in sample buffer and subjected to SDS–PAGE. The blots The tree shows five branches (I–V) indicating the order in which were incubated with either anti-XPA or anti-ERCC1 antibodies (Santa patients were grouped based on similarities in their pattern of gene Cruz Biotechnology, Santa Cruz, CA) and revealed with a peroxidase- expression. (C) Black bar beside the gene dendogram corresponds to conjugated anti-rabbit immunoglobulins (Pierce). the cluster of upregulated genes in cirrhoses. P. Zindy et al. / FEBS Letters 579 (2005) 95–99 97 rect hybridization of the radioactive probe on the target branches clustered fibrotic samples according to their stage cDNA. To further clarify the relationships between the fibrotic (branche III for F1, branche IV for F3 and branche V for stages and genes analyzed, we performed hierarchical cluster- F2). Differences in gene profiles between the five cirrhotic sam- ing of the 14 fibrotic/cirrhotic livers and genes (Fig. 1A). The ples appeared associated with the necroinflammatory activity, resulting dendogram showed five main cluster branches (I–V) since the two cirrhotic livers in branche II showed an enhanced (Fig. 1B), the length of branches reflecting the relatedness of activity (A1 and A3) compared with the three other cirrhotic samples. The two nearest clusters, including branches I and livers included in branche I (A0). When analyzing the cluster- II, contained the five cirrhotic samples and the three other ing of genes based on their relative expression, a major cluster

Table 2 Major gene networks of significantly overexpressed genes in cirrhosis (two-class t test with SAM; FDR = 10%) Symbol Full name ID Mean ratio Inflammatory response CXCL12 Chemokine (C–X–C motif) ligand 12 Hs.237356 2.0 BTG2 BTG family, member 2 Hs.75462 6.8 AIF1 Allograft inflammatory factor 1 Hs.76364 3.6 CSF1R Colony stimulating factor 1 receptor Hs.174142 2.8 VEGF Vascular endothelial growth factor Hs.73793 11.1 IL1A Interleukin 1, a Hs.1722 1.9 IL1B Interleukin 1, b Hs.126256 2.5 TYROBP TYRO protein tyrosine kinase binding protein Hs.9963 3.4

Fibrogenesis COL6A1 Collagen, type VI, a1 Hs.108885 8.0 COL6A2 Collagen, type VI, a2 Hs.420269 2.8 COL6A3 Collagen, type VI, a3 Hs.233240 15.5 COL11A1 Collagen, type XI, a1 Hs.439168 4.8 MMP12 Matrix metalloproteinase 12 Hs.1695 3.0 HSPG2 Heparan sulfate proteoglycan 2 (perlecan) Hs.211573 6.8 DCN Decorin Hs.433989 5.1 TIMP1 Tissue inhibitor of metalloproteinase 1 Hs.5831 3.1 COPEB Core promoter element binding protein, ZF9 Hs.285313 7.0

Cell proliferation NME2 Non-metastatic cells 2, protein (NM23B) expressed in Hs.433416 27.4 RFC4 Replication factor C (activator 1) Hs.35120 4.5 PCNA Proliferating cell nuclear antigen Hs.78996 4.2 RFC3 Replication factor C (activator 1) 3.38 kDa Hs.115474 5.8 CDK6 Cyclin-dependent kinase 6 Hs.38481 4.7 PDGFRA Platelet-derived growth factor receptor, a polypeptide Hs.74615 8.6 PDGFB Platelet-derived growth factor b polypeptide Hs.1976 7.5 MYC c-myc oncogen Hs.202453 9.6

Cell death RAF1 c-raf proto-oncogen Hs.257266 2.8 TNFRSF6 Tumor necrosis factor receptor superfamily, 6 Hs.82359 2.7 CASP4 Caspase 4, apoptosis-related cysteine protease Hs.74122 2.4 TRADD TNFRSF1A-associated via death domain Hs.89862 6.0 BAX BCL2-associated X protein Hs.159428 20.4 BCL2 B-cell CLL/lymphoma 2 Hs.79241 3.1 MCL1 Myeloid cell leukemia sequence 1 (BCL2-related) Hs.86386 4.9 CASP1 Caspase 1, apoptosis-related cysteine protease Hs.2490 9.0 CASP7 Caspase 7, apoptosis-related cysteine protease Hs.9216 2.7 CASP8 Caspase 8, apoptosis-related cysteine protease Hs.381231 6.2

Response to DNA damage CDC34 Cell division cycle 34 Hs.423615 5.3 CHES1 Checkpoint suppressor 1 Hs.211773 2.6 UBE2A Ubiquitin-conjugating enzyme E2A (RAD6 homolog) Hs.379466 38.2 UBL1 Ubiquitin-like 1 (sentrin) Hs.81424 10.2 RAD23A RAD23 homolog A (S. cerevisiae) Hs.180455 6.7 RAD50 RAD50 homolog (S. cerevisiae) Hs.41587 2.1 HMGB2 High-mobility group box 2 Hs.80684 6.2 ERCC3 Excision repair cross-complementing Hs.77929 6.9 DDIT3 DNA-damage-inducible transcript 3 Hs.400353 33.7 DDB2 Damage-specific DNA binding protein 2, 48 kDa Hs.77602 8.0 ERCC1 Excision repair cross-complementing Hs.59544 4.4 XPA Xeroderma pigmentosum, complementation group A Hs.288867 5.0 XRCC1 X-ray repair cross-complementing protein 1 Hs.98493 1.8 ATM Ataxia telangiectasia mutated Hs.526394 4.7 98 P. Zindy et al. / FEBS Letters 579 (2005) 95–99 of upregulated genes in cirrhotic livers was identified (Fig. 1C). TRADD, TNFRSF6/FAS, the caspase family (CASP1, 4, 7 Since hierarchical clustering analysis showed distinct gene and 8) and genes associated with the mitochondrial pathway, expression pattern between cirrhosis and fibrosis, we selected i.e., BAX, BCL2 and MCL1. genes whose expression is significantly altered in cirrhosis by An unexpected finding was the upregulation of genes in- using a two-class t test with permutation using statistical tool volved in DNA repair, including mechanistically distinct repair SAM. GO analysis of these genes revealed gene networks asso- systems [10,11], i.e., nucleotide-excision repair: ERCC3, ciated with molecular pathways previously described in RAD23A, XPA and ERCC1 related to lesions arising from chronic liver injuries, including inflammatory response, fibro- exogenous sources, base-excision repair: XRCC1, homologous genesis, cell proliferation, and cell death (Table 2). Thus, the recombination: ATM and RAD50 which repair double strands increase in chemokine CXCL12 expression (stromal cell- breaks, and the post-replicative repair: UBE2A and UBL1. In derived factor 1), which activates leukocytes in response to a addition, we showed upregulation of CDC34 and ches1, two proinflammatory stimulus, supported the recent report of CXCL12 induction in biliary epithelial cells of inflammatory liver diseases [7]; in addition, TYROBP (DAP12), a transmem- brane adapter molecule for tyrosine kinase receptor, has been involved in liver hepatic granulomatous inflammation [8]. Fibrogenesis hallmarks included collagen VI, perlecan, decorin, TIMP1 and COPEB (ZF9), a DNA-binding protein CPBP which was previously described as an important signal in hepatic stellate cells activation by transactivating promoters driving fibrogenesis, including fibrillar collagens [9]. Cellular proliferation was characterized by an upregulation of PCNA (proliferating cell nuclear antigen) and genes involved in apoptosis included the tumor necrosis factor receptor superfamily, i.e.,

A 10

2 mRNA (Fold increase) P53 XPA ATM UBE2A ERCC1 XRCC1 RAD23A

P53 ATM RAD23A UBE2A ERCC1 XPA

ATM 0.93*** RAD23A 0.79*** 0.79** UBE2A 0.89*** 0.82*** 0.83*** ERCC1 0.73** 0.88*** 0.83*** 0.65* XPA 0.89*** 0.90*** 0.94*** 0.86*** 0.84*** XRCC1 0.84*** 0.91*** 0.79*** 0.70** 0.85*** 0.83***

C 14 12

8 6 ** ** * 4 ** * ** ** 2 mRNA (Fold increase) 0 P53 XPA ATM UBE2A ERCC1 XRCC1 RAD23A Fig. 3. ERCC1 and XPA in cirrhotic livers. (A) Western blotting of Fig. 2. RT-qPCR analysis of DNA repair gene expression in 14 protein extracted from liver sections. (B) Immunochemical staining of cirrhosis. (A) Data are plotted as medians (square) and 25–75th ERCC1 (a) and XPA (b) vascular endothelial cells (v) and biliary percentiles (box). (B) Spearman rank order correlations between epithelial cells (bc) are labeled. ERCC1 and XPA staining is weak or quantified mRNA levels. (C) Association of DNA repair gene absent in control livers (c and d, respectively). Original magnification expression with (black bars) or without (white bars) necroinflamma- (a and b, ·100; c and d, ·50). Immunoperoxidase was counterstained tory activity in cirrhoses (*P < 0.05; **P < 0.01; and ***P < 0.001). with hematoxylin. P. Zindy et al. / FEBS Letters 579 (2005) 95–99 99 genes associated with checkpoint mechanism, a conserved pro- endonuclease activity [16]. Western blots analysis showed that cess in cell division for maintenance of genomic stability which the amounts of both proteins were induced in cirrhotic livers induces DNA strand breaks-inducible cell cycle arrests at G1 (Fig. 3A). Interestingly, immunostaining for ERCC1 and and G2 [12]. Increased expression of XPA, ERCC1, ATM, XPA was predominantly associated with endothelial cells of XRCC1, RAD23A and UBE2A was confirmed by RT-qPCR vessels and biliary epithelial cells (Fig. 3B), thus indicating on a large set of cirrhotic tissues (n = 14) (Fig. 2A). Further- that non-parenchymal cells, in addition to hepatocytes more, all DNA repair gene mRNA levels were highly (Langouet et al., unpublished data), activate DNA repair correlated (Fig. 2B). Although deficient repair of DNA lesion pathways in cirrhosis. O6-methylguanine in cirrhosis was recently reported [13] and a In conclusion, our data show an upregulation of DNA re- 399Gln polymorphism in XRCC1 gene was found associated pair genes in cirrhosis associated with necroinflammatory with a high risk of liver cirrhosis [14], DNA repair activity in activity, suggesting enhanced DNA repair activity to restore chronic liver diseases is poorly documented. In the present higher damages induced by exogenous and endogenous study, overexpression of several DNA repair associated genes DNA reactive agents. might reflect DNA repair perturbation in cirrhosis. Because p53 tumor suppressor activity has been suggested to Acknowledgement: The authors thank Pr. Paulette Bioulac-Sage (Ser- trigger DNA repair in response to various types of stress, we vice dÕAnatomie Pathologique, Hoˆpital Pellegrin, Bordeaux) and Pr. Karim Boudjema and Bernard Meunier (Departement de Chirurgie investigated p53 gene expression. Steady-state p53 mRNA lev- Visce´rale, Rennes) for helpful cooperation. els were significantly increased in cirrhotic livers and Spearman correlation analyses showed strong associations between p53 and DNA repair genes expression, namely ATM (R = 0.93), References RAD23A (R = 0.79), UBE2A (R = 0.89), ERCC1 (R = 0.73), XPA (R = 0.89) and XRCC1 (R = 0.84) (Fig. 2B). Since con- [1] Ferrell, L., Wright, T., Lake, J., Roberts, J. and Ascher, N. (1992) troversial data about increased expression and mutations of Hepatology 16, 1372–1381. p53 gene were reported in cirrhosis, p53 gene sequence was [2] Furuya, K., Nakamura, M., Yamamoto, Y., Togei, K. and investigated in the 14 cirrhotic livers. Double sequencing, per- Otsuka, H. (1988) Cancer 61, 99–105. [3] Paradis, V., Laurendeau, I., Vidaud, M. and Bedossa, P. (1998) formed for exons 4–8 with oligonucleotides designed for DNA Hepatology 28, 953–958. amplification, revealed no missense mutation in the studied [4] Thorgeirsson, S.S. and Grisham, J.W. (2002) Nat. Genet. 31, 339– samples (data not shown). 346. To clarify the upregulation of DNA repair gene network, we [5] Kim, J.W., et al. (2004) Hepatology 39, 518–527. tested a possible association of gene expression with clinical [6] Bedossa, P. and Poynard, T. (1996) Hepatology 24, 289– 293. parameters or pathological features of the cirrhotic livers. [7] Terada, R., et al. (2003) Lab. Invest. 83, 665–672. RT-qPCR data were correlated with sex, age, alcohol, virus [8] Nochi, H., et al. (2003) Am. J. Pathol. 162, 1191–1201. and necroinflammatory activity. Spearman rank order correla- [9] Ratziu, V., Lalazar, A., Wong, L., Dang, Q., Collins, C., tion studies showed association between the necroinflamma- Shaulian, E., Jensen, S. and Friedman, S.L. (1998) Proc. Natl. Acad. Sci. USA 95, 9500–9505. tory activity and the steady-state mRNA levels of DNA [10] Wood, R.D., Mitchell, M., Sgouros, J. and Lindahl, T. (2001) repair genes: ATM (R = 0.65, P < 0.05), RAD23A (R = 0.57, Science 291, 1284–1289. P < 0.05), UBE2A (R = 0.64, P < 0.05) and XPA (R = 0.66, [11] Hoeijmakers, J.H. (2001) Nature 411, 366–374. P < 0.01). In active cirrhosis (n = 9) versus quiescent cirrhoses [12] Dasika, G.K., Lin, S.C., Zhao, S., Sung, P., Tomkinson, A. and (n = 5), all mRNA levels for DNA repair genes were statisti- Lee, E.Y. (1999) Oncogene 18, 7883–7899. [13] Collier, J.D., Guo, K., Burt, A.D., Bassendine, M.F. and Major, cally increased (Fig. 2C). G.N. (1993) Lancet 341, 207–208. We further validated the association between DNA repair [14] Rossit, A.R., Cabral, I.R., Hackel, C., da Silva, R., Froes, gene expression and activity in cirrhosis by examining, at N.D. and Abdel-Rahman, S.Z. (2002) Cancer Lett. 180, 173– the protein level, two selected genes involved in the repair 182. [15] Reardon, J.T. and Sancar, A. (2002) Mol. Cell. Biol. 22, 5938– of covalent bulky helix-distorting DNA lesions: XPA initiates 5945. DNA repair by stabilizing the proteins responsible for DNA [16] de Laat, W.L., Appeldoorn, E., Jaspers, N.G. and Hoeijmakers, incision [15] and ERCC1 plays a direct role in incision by J.H. (1998) J. Biol. Chem. 273, 7835–7842.